born global, gradual global, and their determinants of

Abstract

Applying the duration analysis on 1,959 newly established small and medium-sized Canadian exporting manufacturers, this study compares the survivability of Born Global and Gradual Global firms in the export market. The unique longitudinal (1997-2005) data set that used in this study is constructed by linking multiple administrative data sources from Statistics Canada. With all else held constant, the probability of survival of Born Global is 6 percent lower than Gradual Global firms. After correcting for endogeneity, the probability of survival of Born Global is 15 percent higher, albeit statically insignificant, than Gradual Global firms. Further, the results show that Born Global and Gradual Global firms have notably different advantages and disadvantages while competing in the global export market. These findings have important implications for both academic researchers and policy-makers.

Job Market paper

Born Global, Gradual Global, and their Determinants of Exit from Exporting By Sui Sui

Department of Economics

Carleton University

Ottawa, Ontario, Canada

December 2009

Sui Sui Page 1 of 59

1 Introduction

A Born Global company is also referred to as an International New Venture (INV) or an

International Entrepreneurship, and has been conceptualized as “a company which, from or near

its founding, seeks to derive a substantial proportion of its revenue from the sales of its products

in international markets” (Knight, 1997, p1). The phenomenon of Born Global has become a

popular academic concern, since it challenges the traditional model of internationalization--the

stage model. According to the stage model, the internationalization process of firms is “a process

in which the enterprise gradually increases its international involvement” (Johanson and Vahlne

1990, p. 11).

The purpose of this paper is to contribute to this debate by investigating the effect of

different internationalization processes on the survivability of firms in the export market. There

are two types of internationalization processes that this study compares: Born Global and

Gradual Global. Specifically, a firm is classified as a Born Global if it has started to export

within two years of its inception and that during its first year of export activity, no less than 25%

of its revenue is from exporting; otherwise, a firm is classified as a Gradual Global.

This paper makes three contributions to the literature. First, it contributes compelling

evidence to a growing body of literature in the field of international business and international

trade and has many important implications for international management, entrepreneurship, and

strategic management by seeking to answer the following questions:

(1) What is the impact of the Born Global internationalization process on the survivability of

firms in the export market?

According to a review study by Keupp and Gassmann (2009), there have been at least 179

articles on Born Global published in 16 top-tier journals over the past 14 years. The only study

that examined the survival of Born Globals is written by Mudambi and Zahra in 2007. While

Mudambi and Zahra’s study identified the survivability of Born Globals in the domestic market,


the survivability of Born Globals in the global export market were not investigated. By

investigating this question, this study contributes to the debate between the stage model and the

Born Global phenomenon. For example, the evidence of a negative impact of Born Global on the

survivability of firms in the export market may imply that the gradual global internationalization

process could increase firms’ probability of survival in the global market, and thus, support the

idea of the stage model. The evidence of a positive impact of Born Global on the survivability of

firms in the export market, on the other hand, may imply that Born Global is a strategic choice

for a particular group of young firms. These firms would have a lower probability of survival in

the export market if they have chosen the traditional gradual global internationalization process.

(2) What are the factors that determine Born Globals’ survival in the export market? What

are the factors that determine Gradual Globals’ survival in the export market?

Export activity is an important catalyst for the growth of young firms (Oviatt and McDougall,

1994; Zahra, Ireland, and Hitt, 2000) since the uncertainty and risk of exposing young firms to

foreign markets will trigger their dynamic capability for exploiting new opportunities and

resources (Sapienza et al., 2006). A study by Eaton et al. (2007) on Colombia exporters found

that, only a small fraction of new exporters are able to continuous exporting in the following

year. However, these successful new exporters account for almost half of total export expansion

in less than ten years. Therefore, it is important to identify the factors that determine the success

or failure of new exporters in the international market. Furthermore, by identifying factors that

determine the survival of Born Globals’ in the export market, especially factors that are different

from those that determine the survival of Gradual Globals, the results of this paper enrich our

understanding of the comparative advantages of Born Globals’ in the global export market.

Answering these questions requires quality dynamic firm-level data. After reviewing and

assessing fifty-five empirical studies that were published between 1989 and 2002 within the field

of Born Global, Coviello and Jones (2004) noted that Born Global research is characterized by


static, readily obtainable, and judgment-based data. Among these fifty-five studies, only five

studies are longitudinal and twelve studies used random selection of their sample design. In

contrast, my data set is drawn from three large-scale administrative databases from Statistics

Canada: the Exporter Register (1993-2005), the Business Register (1997-2005), and the

Longitudinal Employment Analysis Program (1997-2004). The Exporter Register, the major

database that is used in this study covers the universe of Canadian exporting firms that have at

least one shipment to a foreign market from 1993 to 2005. Therefore, the second contribution of

this study stems directly from the unique longitudinal firm-level data set I constructed, which

includes the export activities of a relatively large sample of Canadian firms between 1997 and

2005.

Third, in recognition of the possibility that a firm’s choice on its internationalization process

is endogenous, simple comparisons of Born Global and Gradual Global firms are unlikely to

have a causal interpretation on the impact of Born Global on the survivability of firms in the

export market. Therefore, this study applies a two-stage estimation methodology to correct for

the selectivity bias. Furthermore, a reduced-form duration model, the Cox proportional hazard

model, is employed to examine the survivability of firms in the export market. Compared to the

logit model that has been applied in Mudambi and Zahra (2007)’s study, the hazard model that

has been applied in this study has many advantages. Firstly, the logit model can only address the

unconditional probabilities, such as a firm’s probability to exit from exporting in a moment t .

The hazard model, on the other hand, allows me to address the conditional probabilities, such as

a firm’s probability to exit from exporting in a moment t , given that it has continuously

exported until this period t . Secondly, to use the logit model, the researcher needs to select the

time when firms’ characteristics are observed. This selection bias may cause the estimated

probabilities to be biased and inconsistent with the true probabilities. The hazard model, on the

contrary, can produce more efficient and consistent results by accounting for the duration

dependence and using all the available information at each point in time.


The rest of the paper begins with a brief review of the relevant literature and the hypotheses

to be tested in section 2. The data is presented in section 3. Attention is then given to the

methodology in section 4. Section 5 presents empirical results with discussions. Finally,

concluding remarks are discussed in section 6.

2 Theory and Hypotheses

Despite the increase in Born Global studies, neither name nor definition of such groups of

firms has been universally accepted (Keupp and Gassmann, 2009). Such groups of firms have

been given different names such as Born Internationals (Ray, 1989), Born Globals (Rennie,

1993), and International New Ventures (Oviatt & McDougall, 1994). Different conceptual

definitions have also been offered by researchers to describe Born Globals. Examples are: “a

business organization that, from inception, seeks to derive significant competitive advantage

from the use of resources and sale of outputs in multiple countries” by Oviatt and McDougall

(1994, p. 470); firms that “adopt an international or global approach right from their birth or very

shortly thereafter” by Madsen and Servais (1997); “a young entrepreneurial company that

initiates international business activity very early in its evolution, moving rapidly into foreign

markets” by Cavusgil et al. (2008), and “SMEs with accelerated internationalization potential

and global market vision” by Gabrielsson et al. (2008).

Regardless of different definitions, what these firms have in common is their early and fast

approach as opposite to the gradual approach of internationalization process that have been

described by the traditional internationalization model--the stage model. The stage model, or the

Uppsala model, is the earliest internationalization model that was developed by a group of

Swedish economists (Johanson & Wiedersheim-Paul, 1975; Forsgren & Johanson, 1975) who

worked at the Uppsala University. A considerable number of empirical studies have found

evidence that supports this model from different countries. Examples are Johanson and

Wiedersheim-Paul (1975) on Swedish firms, Welch and Luostarinen (1988) on Finnish industrial


companies, and Lam and White (1999) on small companies in the UK. According to the stage

model, firms are cautious and become involve gradually in foreign business activities because of

the risks and uncertainty associated with foreign market entry and survival. Therefore, if the

stage theory is relevant, Born Globals would have a lower chance of survival in the foreign

markets than the Gradual Globals.

Beginning in the early 1990s, however, a number of empirical studies in the have reported an

intriguing phenomenon that contradicted the traditional internationalization theory.1 Brush

(1992) found that 13% of small US manufacturers had started international activities during the

first year of operations. Rennie (1993) observed that about 25% of Australian exporters began

exporting in substantial quantities right from the birth of the company. In their seminal work,

Oviatt and McDougall (1994) argued that the stage model cannot be applied in some cases,

especially among newly established firms who belong to industries that are characterized by

intense international competition and unique knowledge. Similar conclusions have been made by

other empirical studies such as Bell’s (1995) study on computer software companies and

Shrader’s (2001) study on high-technology new ventures.

2.1 The Impact of Born Global on the Survival of New Exporters

Sapienza et al. (2006) developed a dynamic capabilities framework to explain the effect of

Born Global on firm survival and growth. Based on this theoretical framework, Born Global

firms may experience lower survival rate than Gradual Global firms because they need time to

develop routine organizational processes and positional advantages in the foreign market. A

study by Zahra (2005) made the same prediction, that Born Globals would have a relatively

lower probability of survival because of their disadvantage in organizational age, prior

1 Studies by Zahra (2005) and Meckl and Schramm (2005) are the two recent reviews on the theory and empirical

evidence of Born Globals.


managerial experience, and resource constraints.

In the Industrial Organization literature, one of the stylized facts is that age and size are

positively related to the probability of a firm’s survival (Geroski, 1995). For example, decreasing

hazard rates with age has been found among US firms (Audretsch and Mahmood, 1995),

Canadian firms (Baldwin and Gorecki, 1991), and Portuguese firms (Mata et al., 1995). New

firms need time to develop their new organizational capabilities and face a high probability of

exit (Stinchcombe, 1965). The age-effects on survival, or the so-called liability of ‘‘newness’’2,

refer to the time firms need to establish themselves, carry out specific investments, develop

knowledge and appropriate production routines, and build up with business partners. Based on

this argument, by choosing to become international in a slow and incremental manner, Gradual

Global firms could have a better chance of survival in the foreign markets by gaining knowledge

and experience and building up sufficient capacity for selling their products worldwide over

time.

In the field of International Entrepreneurship, Born Global is described as “a combination of

innovative, proactive, and risking behavior that crosses national boundaries and is intended to

create value in organizations” (McDougall & Oviatt, 2000, p903). Younger firms, as reported by

Ursic and Czinkota (1984), often are lacking of cost advantages, sufficient amount and source of

resources and financial assistant to help them compete in the local market, hence are more

interested in the global market than the older ones. Conditional on “improvements in global

telecommunications and transport networks, combined with increasingly liberalized global

trading regimes” (Fan and Phan, 2007, p1113), Born Global is a strategic choice for a particular

group of firms. For example, the founders of these companies may be familiar with foreign

markets because of their prior international business experiences (Knight and Cavusgil, 1996;

2 The terminology of the liability of newness is original from Freeman et al. (1983).


Oviatte and McDougall, 1997). In this case, a firm’s decision on its internationalization process

is endogenous. Consequently, one has to take into account for the endogeneity of a firm’s choice

on its internationalization process when assessing the impact of Born Global on the survivability

of firms in the export market.

To the best of the author’s knowledge, the impact of Born Global on the survival of firms in

the export market has not been investigated empirically. The existing empirical evidence on the

survival of Born Globals is not entirely conclusive either. A study by Mudambi and Zahra (2007)

found that Born Globals are significantly less likely to survive than established firms. However,

the lower probability of survival associated with Born Globals disappears when the endogeneity

on a firm’s choice on its internationalization process is taken into account. The results of their

study suggest that early internationalization maybe a value-maximizing organizational form and

an endogenous optimal strategy for a particular group of firms. Empirical models that do not

account for self-selection of strategy choice, suggested by Shaver (1998), can be misspecified,

and lead to incorrect conclusions.

The theoretical considerations described above yield two hypotheses with regard to the

effect of the Born Global internationalization process on the survivability of firms in the export

market, which I will explore empirically:

Hypothesis 1a: Everything else being equal, Born Globals have a higher probability of exit

from exporting than Gradual Globals.

Hypothesis 1b: After endogenizing a firm’s choice on its internationalization process,

Born Globals have a probability of exit from exporting no greater than Gradual Globals.

2.2 Exit from Exporting of Born Global and Gradual Global Firms

Productivity. In the international trade literature, studies from the microeconometrics of


international firm activities have noticed the superior performance of exporters relative to

non-exporters (e.g. Bernard and Jensen, 1999; Isgut, 2001, Clerides et al., 1998; Delgado et al.,

2002). For example, firm-level evidence from European countries (Mayer and Ottaviano, 2008)

revealed that internationalized firms are “superstars”, they “are bigger, generate higher value

added, pay higher wages, employ more capital per worker and more skilled workers, and have

higher productivity”.

An important theoretical framework to explain such findings is the Melitz (2003) model with

heterogeneous firms. The Melitz model is a theoretical model that incorporates firm

heterogeneity into the Krugman’s (1979) monopolistic competition framework. In that model,

there are fixed costs and variable costs associated with entering the industry and entering the

export market. As a result, only a small fraction of productive firms engage and remain in

exporting. Furthermore, wage is used as a proxy for human capital intensity by Wagner (2003) to

reflect different levels of labor productivity. Based on Wagner’s argument, one would expect

firms that pay higher wage rates to be more productive, and thus, more likely to survive in the

export market. The least productive firms are forced to exit. These arguments lead to the

following hypotheses:

Hypothesis 2a: Everything else being equal, firms that produce more revenue per worker

have higher probability of survival in the export market.

Hypothesis 2b: Everything else being equal, firms that pay higher wage rates have higher

probability of survival in the export market.

Export Market Diversification. Geographic concentration of export destinations leaves

exporters vulnerable in case of rapid changes in the political or economic situations of their key

trade partners. By exporting to multiple destinations (especially countries with different business

cycles, exchange rate regimes, etc.) exporters are less likely to run the risk of depending on one

trading partner, and thus, can maintain a certain level of threshold to be profitable. Hence, more


market diversified exporters are more likely to survive in the international market. Similar

arguments can be applied to product diversification. In other words, exporting to more

destinations and exporting more varieties of products could provide the insurance benefit to

exporters because the risks of uncertain demand are less correlated from different types of

markets and products. Furthermore, a study by Eaton et al. (2005) suggested that more

productive firms should be able to export to more destinations. Consequently, more diversified

firms are more productive, and are more likely to survive in the export market. These arguments

lead to the following hypothesis:

Hypothesis 3a: Everything else being equal, firms that are more market-diversified are more

likely to survive in the export market.

Hypothesis 3b: Everything else being equal, firms that are more product-diversified are

more likely to survive in the export market.

US first. A study by Sabuhoro et al. (2006) suggested that for Canadian firms, choosing the

US as the first export destination has a positive impact on their probability of survival in the

export market. This result is consistent with the idea of the stage model, that after building strong

appearances in the domestic market, firms often export first to a single neighboring market and

subsequently enter more distanced foreign markets (Bell, 1995). This finding is also consistent

with a study conducted by Eaton et al. (2007), who investigated the cross market dynamics of

Colombian firms from 1996 to 2005. The authors showed that it appears to be a popular

geographical expansion strategy for Colombian exporters to present themselves gradually from a

single neighboring foreign market to additional destinations within the regional market, and then

reach to larger OECD markets. Eaton et al. suggested that the likelihood of the survival of

exporters depends on their choice of initial market destination since success in neighboring

markets may suggest that the expected payoff could exceed the sunk cost in more distanced

markets.


When it comes to Born Global firms, however, physical distance is argued to be less

important during their internationalization process (Boter and Holmquist, 1996; Keeble et al.,

1998; Madsen et al., 2000). These arguments lead to the following hypotheses:

Hypothesis 4a: Everything else being equal, choosing the US as the first export destination

has a positive effect on the survivability of Gradual Global firms.

Hypothesis 4b: Everything else being equal, choosing the US as the first export destination

does not affect the survivability of Born Global firms.

3 Data

3.1 Data Sources

The data I constructed to examine my research questions gathers information from the

Exporter Register (ER), the Business Register (BR), and the Longitudinal Employment Analysis

Program (LEAP). All these databases are produced and maintained by Statistics Canada. The

creation of the analysis data set involved the aggregation of establishment-level data to the

enterprise level3, the generation of the variables of interest from administrative records, the

aggregation of annual data to panel data, and the merge of different data bases.

I constructed my data set at the enterprise level. Any reference made to “exporter” or “firm”

in this study represents a “statistical enterprise”. A unique and static identification number (id) is

assigned to each firm in each database used in this study. Using this firm id as a time-invariant

identifier, I linked the original annual ER database and built a panel that goes from 1993 to 2005.

The LEAP panel goes from 1997 to 2004 and the BR panel goes from 1997 to 2005. I linked

these three panels and built the final data set used in this study.

3 An enterprise may have more than one establishment. For more detailed scandalized classification of both

enterprise and establishment, please see Page 8 of “A Profile of Canadian Exporters, 1993-2005”.


My main data source, the Exporter Register (ER), is a large-scale administrative database of

all merchandise trade transactions by Canadian firms at both the establishment and enterprise

level from 1993 to 2005. The data was obtained from two sources: the US Customs documents

and the Canada Revenue Agency (CRA) documents. ER is produced and maintained by the

International Trade Division at Statistics Canada. Each transaction is recorded separately in ER. I

aggregated transactions by a given firm to obtain its total value of exports, its number of export

destinations, and the variety of products it exports in each year between 1993 and 2005. A

transaction record includes the firm’s identification number, a product code that is classified at

an 8-digit Harmonized Schedule (HS8), the value of transaction in Canadian dollars, and the

country of destination. If the country of destination is the US, information on the destination

state is also recorded. If a firm did not export, or its total value of export is $2000 Canadian

dollars or less in a given year, this firm is not included in the ER for that year.

The ER database also provides information with regard to the industry to which a firm is

classified from a 6-digit North American Industry Classification System (NAICS6) code and the

province in which a firm is located. This data set allows me to track the entry and exit of firms in

the foreign markets, the number of years a firm exported between 1993 and 2005, its value of

exports, the destinations and the products it exports in each year between 1993 and 2005.

The second data source, the Business Register (BR), is a main frame that includes essentially

all businesses operating within Canada as well as foreign businesses that have links with

Canadian companies from 1987 to 2006. BR is maintained by the Business Register Division

(BRD) at Statistics Canada. It contains a complete, up to date and unduplicated list of businesses

that have a corporate income tax (T2) account, are an employer, or have a GST account. I used

the BR database as supplements for the ER database for information on firms’ annual revenue,

country of ownership, and first year of business.


The third data source, the Longitudinal Employment Analysis Program (LEAP), contains

employment and payroll information for each employer business in Canada and is available at

the enterprise level. LEAP is maintained by the Science, Innovation and Electronic Information

Division (SIEID) at Statistics Canada and is available between 1997 and 2004. In this data set,

each observation corresponds to a unique combination of

year-enterprise-province-payroll-individual labor units (ILU). If an enterprise has plants located

in different provinces, its corresponding information is available at the provincial level. The

variable ‘payroll’ measures total payroll each enterprise pays in each province in a given year.

The variable ‘ILU’ measures the labor units an enterprise hires in each province in a given year.

In most cases, an ILU represents one employee. In cases where one person works for several

companies in a year, his or her ILU is distributed proportionally among these enterprises. A

part-time worker would still contribute one ILU to the total if he or she did not work for another

enterprise.

3.2 Sample Selection

I now present the criteria and procedure of selecting observations. A study on the conceptual

definition of Born Global by Gabrielsson et al. (2008) emphasized that Born Global firms should

be “small and medium-sized enterprises (SMEs) with accelerated internationalization potential

and global market vision”, and when it comes to defining Born Globals, one should take into

account “factors such as the type of industry”. Following Gabrielsson et al., I selected the

suitable observations for this study using the following criteria:

First, I selected firms that belong to the manufacturing industry. I decided to focus my

analysis on the manufacturing industry because this study is intended to investigate firms that

manufacture their own products. In other words, I want to eliminate Intermediation firms that sell

products produced by other companies, such as, firms belonging to the wholesale trade industry.

As shown in Figure 1, the initial data set of ER contains 113,113 enterprises across all industries

that have at least one shipment to any foreign market between 1993 and 2005. Each enterprise


has a corresponding 6-digit North American Industry Classification System (NAICS6) code.

Based on this code, I have information on the industry to which each enterprise belongs. I

selected 34,659 out of these 113,113 (30.64%) enterprises that belong to the manufacturing

industry for the purposes of this study.

Figure 1. Sample selection

Second, I selected firms that were established between 1997 and 2004. This study uses the

founding condition to classify a firm as Born Global or Gradual Global. Because information on

firm founding conditions from the LEAP database is only available between 1997 and 2004,

firms that were established prior to 1997 were excluded. Among the 113,113 enterprises that

appear in the initial ER data set, 62,461 (55.22%) enterprises appear in the BR database. Many of

these enterprises that appear in the ER database but not in the BR database are unincorporated

businesses, individuals, or institutions whose export behaviors are very irregular. I used the BR

database as the main frame of my data set. This implies that enterprises that appear in the ER but


not the BR database are not considered as firms, and thus, were excluded in this study. For each

enterprise, BR provides information on the year in which the enterprise entered the BR database.

I selected 26,603 out of these 62,461 (42.59%) enterprises that entered the BR database between

1997 and 2004 for the purposes of this study.

Third, I selected firms with less than 500 employees because the aim of this study is to

investigate the internationalization process of small and medium-sized enterprises (SMEs). Firms

that do not appear in the LEAP database were excluded. Among the 113,113 enterprises that

appear in the initial ER data set, 43,691 (38.63%) of them appear in the LEAP database. Among

these 43,691 enterprises, 41,903 (95.91%) enterprises with less than 500 employees (measured

by individual labor unit) were selected for the purposes of this study.

Using the unique identification number assigned to each enterprise, I linked the ER, BR, and

LEAP databases. My sample is reduced to 3,061 enterprises. I cleaned the sample and excluded

outliers by using the following criteria: first, in order to make sure that firms in my sample are

active, I eliminated 662 enterprises whose annual revenue is never higher than $30,000 Canadian

dollars in my research period (remaining sample 2,399). Second, in order to make sure that

exporting is an important part of a firm’s business activity, I eliminated 420 enterprises whose

annual value of exports is never higher than $2,000 in my research period (remaining sample

1,979). I found that some values of the variable ‘payroll per worker’ in my sample are

unreasonably high or low. It is possible that the values of this variable were incorrectly entered.

Therefore, I eliminated the highest and the lowest 0.25% outliers to avoid possible misleading

results. This gives the analysis sample of 1,959 enterprises.

The method of trend imputation is applied in constructing my data set. For example, the

enterprise-level employment and payroll information is available between 1997 and 2004, but is

not available in 2005. I imputed an enterprise 2005’s employee and payroll from its previous

years’ information with a year-to-year change ratio where there are indications that the enterprise

has activity in 2005.


3.3 Construction of the Variables of Interest

In Appendix A, Table 1 presents definitions, data sources, and descriptive statistics for the

variables used in this paper. I constructed the following variables in order to classify the Born

Global firm. ‘BRBY’ (Business Register birth year) is estimated by the first year a firm appears

in the BR database, which is used as a proxy for the first year a firm starts its business. Since the

BR database starts in 1987, this information is unknown for those firms that are established

before 1987. Accordingly, I generated a set of dummy variables ‘Cohort’, which equal one if a

firm starts its business in a particular year. An exporter may register in the ER before it registers

in the BR database. Such exporters are usually individuals who have to register the ER database

because they need to ship products to a foreign buyer. They do not have to register in the BR

database unless they want to form a company. Therefore, such firm is not considered as a firm

until it registers in the BR database in this study. ‘ERBY’ (Exporter Register birth year) is

estimated by the first year a firm registers in the ER database after it registered the BR database,

which is used as a proxy for the first year a firm starts to export.

A firm’s export start-up age, ‘ExAge’, is measured by the difference between a firm’s ER

birth year and BR birth year plus one. If a firm enters the BR database and the ER database in the

same year, its start-up age equals to one. If a firm enters the ER database before it enters the BR

database, its export start-up age equals to one as well. Enterprises that do not register in the BR

database are not considered as firms, and thus, were excluded in my data set.

The variable ‘Exports’ is measured in millions of Canadian dollars from the ER database

and is deflated by annual merchandise exports customs-based price indexes, using base year

2000. The Variable ‘Exports’ is available annually from 1993 to 2005. ‘Revenue’ is a proxy for

the size of a firm measured in millions of Canadian dollars from the BR database and is deflated

by annual industry price indexes, using base year 2000. The Variable ‘Revenue’ is available

annually from 1997 to 2005. A firm’s export intensity, ‘EI’, is measured as the ratio of export

sales to revenue of the firm and is available annually from 1997 to 2005. I found some firms


have higher level of exports than revenue. One possibility is that the moment a firm reports its

revenue might be different than the moment it reports its exports, which is referred to as a

calendar issue. Another possibility is that either the value of exports or the value of revenue, or

both are incorrectly entered. I let the export intensity of such enterprise equal to one.

‘BG’ is a dummy variable that equals to one if a firm starts to export within two years of its

inception and has exported no less than 25% of its total revenue during the first year of its export

activity. Accordingly, 497 out of 1959 (25.37%) firms are classified as Born Global in this study.

I constructed the following variables that might influence a firm’s survivability in the export

market. Based on a firm’s country of control from the BR database, I generated a dummy

variable ‘Foreign’, which equals to one if a firm’s home country of owner is not Canada. The

percentage of foreign-owned firms in my sample is 1.68%. Based on a firm’s corresponding

5-digit North American Industry Classification System (NAICS5) code from the ER database, I

constructed a dummy variable ‘ICT’, which equals to one if a firm belongs to the information

and communication technology (ICT) sector. In general, the ICT sector is defined as the

combination of manufacturing and service industries that electronically capture, transmit, and

display data and information (Statistics Canada, 2001). The percentage of ICT firms in my

sample is 7.5%. Table B in Appendix B presents the corresponding NAICS5 codes that are

classified as ICT sectors. This classification was developed by Statistics Canada and Industry

Canada in 1999.

From the ER database, I know the first year each firm enters each market. I generated a

dummy variable ‘US first’, which equals to one if a firm exports to the US before it exports to

another country. If a firm only exported to the US market, it is considered as having chosen the

US as its first destination and is included in the ‘US first’ category. The percentage of firms that

choose the US as their first export destination in my sample is 81.98%. This result suggests that

newly established small and medium-sized Canadian manufactures are highly dependent on the

US market.


Based on a firm’s corresponding NAICS3 code from the ER database, I generated a set of

dummy variables ‘Sector’, which are set at one if a firm belongs to a particular sector. There are

eleven sector groups in this study. Based on a firm’s corresponding 2-digit province-of-location

categorical variable from the ER database, I generated a set of dummy variables ‘Province’,

which equal one if a firm is located in a particular province. If a firm has plants in multiple

provinces, I used the province in which the firm derives the highest share of it value of exports as

its province of location.

‘Employees’ is measured by the individual labor unit (ILU) and is available annually from

1997 to 2005 from the LEAP database. As I described in the earlier section, an ILU represents

one employee in most cases. In cases where one person works for several companies in a year,

his or her individual labor unit is distributed proportionally among the enterprises. A part-time

worker would still contribute one ILU to the total if he or she did not work for another enterprise.

The variable ‘Labor Productivity’ is calculated by the ratio of revenue to the number of

workers and is available annually from 1997 to 2005. It is measured in thousands of Canadian

dollars and is deflated by annual industry price indexes, using base year 2000. The variable

‘Wage Rates’ is calculated by the ratio of total payroll to the number of workers and is available

annually from 1997 to 2005. It is measured in thousands of Canadian dollars and is deflated by

annual consumer price indexes, using base year 2000.

The variable ‘Products’ refers to the variety of products a firm exported based on the count

of its HS8 code from the ER database, which is used as a proxy for a firm’s degree of product

diversification. The variable ‘Products’ is available annually from 1993 to 2005. The variable

‘Destinations’ is measured by the number of countries to which a firm exported from the ER

database, which is used as a proxy for a firm’s degree of destination diversification. The Variable

‘Destinations’ is available annually from 1993 to 2005.

I constructed the following variables for the purpose of the duration analysis. The dummy

variable ‘Censor’ is set at one if a firm is exporting in the year which the period of observation


is terminated. 1395 out of 1959 (71.21%) firms in my sample were still exporting in 2005, and

thus, their corresponding value of variable ‘Censor’ equals to one.

Most studies on the exit and entry of exporters consider a firm as having stopped from

exporting if it stays out of the export market for one year or longer. To account for the

seasonality in firms’ export behavior, the following estimation convention is adopted in this

study: a firm is considered as having stopped from exporting if it stays out of export market for

two consecutive years or longer. Based on this methodology, the variable ‘Duration’ is defined

as the number of years a firm remains active in the export market. There are 81 out of 1959

(4.13%) firms that export over the whole period (1997-2005) in my sample, and thus, their

corresponding export duration is nine years. Moreover, to estimate the impact of previous exits

on survival time, I generated a dummy variable ‘Re-enter’ that is set at one if a firm reenters the

export market.

4 Empirical Methodology

4.1 Duration Analysis

This study uses a reduced-form model to identify the determinants of the ability of Born

Globals’ to remain active in the export market. The statistical software I used for my analysis is

STATA 10. My aim is to identify the factors that affect a new exporter’s instantaneous

probability of exit from exporting.

In the duration analysis, the variable of interest is the length of uninterrupted time that elapsed

between a firm’s entry and exit in the export market. Let T denote the duration of exports and

t denote a particular value of T . The survivor function )(tS , the probability that the duration

of the spell T equals or exceeds the value t , is defined as follows:

∫∞

−==≥=t

tFdxxftTPtS )(1)()()( (1)


where (.)f is the probability density function (pdf) and (.)F is the cumulative distribution

function (CDF) for T .

Following Lancaster (1990, chapter 2), the hazard rate of firm i is defined as the

instantaneous probability of i exiting the global market in a moment t , given that it has

continuously exported until this period t , and conditional on a vector of covariates itX , which

includes both time-invariant and time-varying variables,

hxtThtTtPxt iiii

hiti),|(lim);(

0

≥+<≤=

→λ

(2)

where T is specified as a non-negative random variable. I assume that T is a continuous

variable, so that for each t , )( itλ is the instantaneous exit rate for firm i .

This study uses the semi-parametric Cox proportional hazard model (CPHM) specification.

CPHM is flexible in the specification of the baseline hazard )(0 itλ , allows for a proportional

specification for unobserved heterogeneity, and a function of observables. The proportional

hazard model with time-varying covariates was first introduced by Cox in 1972. A particular

advantage of the Cox model is that the baseline hazard is left unspecified and is not estimated.

The basic proportional hazard model takes the form of:

)...exp()();( 110 kkiii xxtxt ββλλ ++⋅= (3)

where )(0 itλ is the baseline hazard for firm i, which measures the hazard of an individual firm

for whom all covariates are zero at a given time t . I introduce a transformation function )( itg

into above regression as the following:

{ })...)((...exp)();( 11110 mmikkiiti zztgxxtxt γγββλλ +++++⋅= (4)


where variables ),...,( 1 mzz are independent of time. However, the transformation function

)( itg makes )...)(( 11 mmi zztg γγ ++ vary continuously with time.

4.2 Counting Process Approach for Multiple Failure Events

The event of exit from exporting may occur more than once for a given firm during the

research period, since a firm may enter, exit, and then re-enter the export market. Such events are

called “multiple events”, or “recurrent events”. To model this type of event, I used the CPHM

and converted the multi-failure event data to a replicated-process single-failure event data, by

assigning a new identification number to a firm if it re-enters the export market. This approach is

called the Counting Process Approach (Andersen et al., 1993). Alternatively, I can restrict the

analysis to the first event of the multiple events only. The drawback of such approach is that it

cannot estimate the impact of a past event on the occurrences of a new event.

This study focused on the Counting Process Approach, but also assigned a set of dummy

variables to each firm to indicate the number of times a firm enters the export market. As I

mentioned in the data section, a firm is considered as having stopped exporting if it stays out of

the export market for two consecutive years or longer. Based on this estimation method, the

maximum number of times a firm could re-enter the export market between 1997 and 2005 is

three times. My data set has an observation of 1,959 firms. The numbers of firms that are entered

the export market once, twice, and three times are 1,395 (71.21%), 112 (5.72%), and one

(0.05%), respectively. In all, my data set includes 2,072 single events.

This length of time is measured in years, which can be either fully observed as a complete

spell, or partially observed as an incomplete spell due to the fact that some firms have an export

duration that pass the observed period. The second type of spell is defined as a censored spell in

the context of duration analysis. More specifically, a spell is left censored if the firm exported in

the first sample year. To minimize the left censor problem, this study only focused on new

exporters. A spell is right censored if the firm is still exporting in the last sample year. My data

set has an observation of 1,959 firms, of which 1,395 are right censored (71.21%).


4.3 Two-Stage Estimation Methodology

The aim of this study is to investigate the impact of Born Global on the survivability of firms

in the export market. In general, the indicator for those who choose the Born Global

internationalization process is endogenous, because the firms who choose this strategy perceive

this strategy beneficial for their long-run profit or export market survivability. In recognition of

possible endogeneity of a firm’s choice on its internationalization process (Masten, 1993), I used

a two-stage estimation methodology that is similar to Mudambi and Zahra’s (2007).

At the first stage, I use a probit model to formulate a firm’s internationalization process

decision between Born Global and Gradual Global4. Because a firm’s decision on whether or not

to choose the Born Global internationalization process is made at the founding of the company,

the setting of this model is cross-sectional.

In general, when a firm’s expected value of return in choosing the Born Global is greater

than its expected value of return in choosing the Gradual Global internationalization process, the

firm will choose the Born Global internationalization process. Otherwise, the firm will choose

the Gradual Global internationalization process.

I defined the variable *iBG as the difference in the expected value of return between the

Born Global and Gradual Global internationalization processes for a new firm i . This value, and

hence the chosen internationalization process, is a function of measureable firm attributes and

cohort conditions and a disturbance term. I cannot observe *iBG but I can observe the chosen

internationalization process ( iBG =1 if Born Global, iBG =0 if Gradual Global) and thus infer

4 The discussion is drawn from Shaver’s (1998)’s empirical model on a firm’s foreign direct investment entry decision between acquisition and greenfield.


whether or not *iBG is positive or negative. This is the standard formulation of a dichotomous

choice model, which can be represented by the following equation:

iiji vYBG +=α* , iBG =1 if 0* >iBG , and 0 otherwise. (5)

where iY is a vector of independent variables (which includes a constant term), jα are the

coefficients I estimate. I assumed iv is normally distributed with zero means and unit variance.

Furthermore, iv represents additional unobserved effects that might affect a firm’s choice

between the Born Global and Gradual Global internationalization processes.

More specifically, BG is the dependent variable in the binary choice model. In this study,

BG equals to one if a firm starts to export within three years of its inception and has exported no

less than 25% of its total revenue during the first year of its export activity. The vector of

independent variables includes initial firm conditions (number of employees, revenue per worker,

and payroll per worker), iICT (a dummy variable and which equals to one if a firm belongs to

the ICT sector), iF (a dummy variable which equals to one if a firm is foreign-owned), iS , iP ,

and iC (sector-, province- and cohort- specific dummy variables respectively).

Accordingly, I estimate iBG^

, a firm’s predicted probability of choosing the Born Global

internationalization process conditional on iY :

iji YBG '^

α̂= (6)

At the second stage, I used the CPHM to estimate the impact of Born Global on the survival

of firms in the export market. I have discussed the CPH model in detail in Section 4.1. In this

section, to emphasize the importance of accounting for the endogeneity, I use a simplified

specification as the following:


iiii BGXSurvival ξηβ ++=^

' (7)

where iSurvival is the probability of survive of firm i , iX is a vector of independent

variables that may affect the survival of the firm, which includes initial firm conditions (number

of employees, revenue per worker, and payroll per worker), current firm conditions (number of

employees, revenue per worker, payroll per worker, variety of products exported, and number of

export destinations), iUSfirst (a dummy variable which equals to one if a firm chose the US as

its first export destination), iICT (a dummy variable and which equals to one if a firm belongs

to the ICT sector), iF (a dummy variable which equals to one if a firm is foreign-owned),

iS , iP , and iY (sector-, province- and year- specific dummy variables, respectively). iBG^

is

the estimated probability for firm i to choose the Born Global internationalization process,

and iξ is an error term that is normally distributed with zero means and unit variance.

Why could the endogeneity of firms’ internationalization process choice lead to biased

coefficient estimates? Suppose I estimate Equation (7) without accounting for the endogeneity on

a firm’s choice on its internationalization process with the following equation:

iiii BGXSurvival εδβ ++= ' (8)

where iBG is a binary variable that indicate whether the internationalization process of firm i

is Born Global or not, and iε is an error term that is normally distributed with zero means and

unit variance. In addition, iε represents the unobserved effects that may influence the survival of

firm in the export market that are not included in iX .

If unobserved effects captured in iε are the same as in iv , these two error terms will be

correlated. For example, prior international business experience of the founder of the company

may increase the likelihood of the firm to choose the Born Global internationalization process, as


well as increase the likelihood for the firm to survive in the export market. Since such effect is

not captured in either iY or iX , it is captured in iv and iε . Therefore, iv and iε are

correlated. In this case, the estimation of δ will not have desirable statistical properties, and the

use of δ to interpret the effect of Born Global on the survival of firms in the export market

would likely be misleading (Shaver, 1998).

4.4 Split-Sample Instrumental Variable Estimation

The method of the Split-Sample Instrumental Variable (SSIV) is applied to check for the

robustness of the results. According to Angrist and Kruege (1995), unlike conventional IV

estimates, SSIV estimates are biased toward zero regardless of the degree of covariance between

structural and reduced-form errors or the first-stage 2R . An unbiased estimate of the attenuation

bias of SSIV is given by the coefficient from a regression of the endogenous regressor on its

predicted value (using data from one half of the sample but using first-stage parameters from the

other half).

Accordingly, I randomly split the sample in half and use one half of the sample to estimate

parameters of the first-stage (binary choice model) equation. These estimated first-stage

parameters are then used to construct fitted values for the endogenous regressor from data in the

other half of the sample. After that, the predicted values of the endogenous regressor are used in

the second-stage (duration analysis) parameter estimates.

4.5 Separated Hazard Function for Born Globals and Gradual Globals

The second aim of this study is to identify the factors that are related to the long-run survival

of Born Globals and Gradual Globals in the export market. Until now, I assumed that the

survivability of exporting of Born Global and Gradual Global firms is explained by the same

factors, and these factors differ only by a proportional factor. However, such assumptions may


discard the possibility that the survivability of exporting of Born Global and Gradual Global

firms is determined by different factors. One approach to overcome such limitations is to

estimate the model separately for Born Global and Gradual Global firms. This approach would

not enforce a common baseline hazard function across the two groups, nor equal effects of the

independent variables.

For above reasons, I ran separate regressions for Born Globals and Gradual Globals. I

control for heterogeneity among firms by including variables that are expected to affect the

survival of firms in the export market: the initial firm conditions such as are initial firm size,

labor productivity, and wage rates; the time-invariant firm characteristics such as ICT,

foreign-owned, and US-first; the time-varying firm specific characteristics such as the present

firm size, labor productivity, wage rates, number of export destinations, and variety of products

exported. The inclusion of sector-specific dummy variables is important to capture the industry

variations in unobserved factors in the data. For the same reason, it is important to include

province-specific and cohort-specific dummy variables.

5 Results

5.1 Classification of Firms

Although there are many different numerical classifications of Born Globals, the common

criteria that have been used to classify Born Globals are start-up age (age of the firm when it

started to export) and export intensity (proportion of revenue that are exported). Period of time a

firm must internationalize to be considered as a Born Global ranges from two years by

McKinsey and Co. (1993) to eight years by McDougall et al. (1994). It is not precise in the

literature on the proportion of revenue that must be exported for a firm to be considered as a

Born Global. The minimum export intensity a firm must achieve to be considered as a Born

Global ranges from 5% by Zahra et al. (2000) to 80% by Chetty and Campbell-Hunt (2004).


Most studies (Rennie, 1993; Moen and Servais, 2002; Madsen et al., 2000) have used the criteria

of 25% that is proposed by Knight and Cavusgil (1996).

Considering the entry cohort of firms (1997-2004) and the research period (1997-2005) of

my data set, I decided to specify a firm as a Born Global if it has started to export within two

years of its inception and have exported no less than 25% of its revenue during the first year of

its export activity. The rest of the firms in my sample are classified as Gradual Globals.

Accordingly to this classification, the number of firms is classified as Born Global is 497

(25.37%), and the number of firms is classified as Gradual Global is 1,462.

Table 1. Comparison of the Initial Conditions between Born Global and Gradual Global firms

Variables Born Global Gradual Global t-test Initial Size = )log( 1revenue -1.25 (1.17) -0.84 (1.08) 0.0000

= )log( 1employees 1.87 (0.05) 2.41 (0.03) 0.0000

Initial Labor Productivity

= )ker/log( 1worrevenue 10.01 (0.67) 11.13 (0.65) 0.0000

Initial Wage Rates

= )ker/log( 1worpayroll 3.10 (0.50) 3.20 (0.47) 0.0000

Initial Export Market Commitment

= )log(exp eorts -1.72 (1.33) -3.09 (1.37) 0.0000

= )log( ensdestinatio 0.13 (0.39) 0.09 (0.31) 0.0082

= )log( eproducts 0.79 (0.74) 0.52 (0.64) 0.0000

ICT (%) 6.84 (25.27) 7.73 (26.71) 0.2582 Foreign-owned (%) 1.21 (10.93) 1.85 (13.47) 0.1649

Note:

a. Value of means is reported in this table, value of standard deviations is reported in parentheses, and value of two-tail t-test, Pr(|T|>|t|), is reported in the last column.


b. Variables ‘employees’, ‘payroll per worker’, and ‘revenue per worker’ are estimated in the first year a firm started its business; variables ‘exports’, ‘products’, and ‘destinations’ are estimated in the first year a firm started to export.

Data Source: the Exporter Register, the Business Register, and the LEAP from Statistics Canada.

In Table 1, the student’s t-test is used to compare the mean of the variable of interest for

Born Global and Gradual Global firms. The distribution of most variables that are used in this

study, such as the variety of exported products and the number of exporting destinations is

substantially skewed towards small numbers. For this reason, I use the natural logs rather than

the original raw values of the continuous variables in my analyses.

Results from Table 1 suggest that Born Global and Gradual Global firms are associated with

very different attributes. Compared to Gradual Global firms, Born Global firms (1) are smaller

both in terms of number of employees and revenues, (2) have lower revenue per worker, (3) have

lower payroll per worker, (4) are more committed to the export market since they have higher

value of exports, number of export destinations, and variety of exported products, and (5) are not

significantly different when it comes to the probability of being foreign-owned or belonging to

the ICT sector.

Will the attributes of the Born Globals change if I adopted different numerical thresholds?

For example, I could specify a firm as a Born Global if it has started to export within two years

of its inception and have exported no less than 50% of its revenue. To address above, Appendix

B delivers a detailed investigation on the attributes of firms that are grouped by different

thresholds of export start-up age and initial export intensity.

In Table B1, firms are grouped by their export start-up age and initial export intensity as in

Table 1. However, both of these two measurements are divided into five rather than two

sub-groups. Specifically, the measurement ‘export start-up age’ is grouped into zero year old,

one year old, two years old, three years old, and four or more than four years old; the

measurement ‘initial export intensity’ is grouped into 50%-100%, 25%-50%, 12.5%-25%,


6.25%-12.5%, and 0%-6.25%. Preferably, I would like to group firms’ initial export intensity

into smaller groups, such as, 90%-100%, 80%-90%, … , 10%-20%, and 0%-10%. Due to

confidentiality issues with regard to the use of data by Statistics Canada, I selected the above

numerical thresholds so that the minimum number of firms in each group is ten (see Table B1 for

the number of firms in each group). The results from Table B2-B10 are qualitatively similar to

my previous results in Table 1. Therefore, my conclusion is that the attributes of the Born Global

and Gradual Global firms will not be drastically affected by adopting different numerical

thresholds.

5.2 The Founding Conditions of Born Global and Gradual Global Firms

Table 2 presents key results for the internationalization process choice model, estimated by a

probit model5. The details of the estimated results are reported in Table 2b in Appendix C. The

coefficients in Table 2 show the effects of the explanatory variables on the marginal utility of the

Born Global relative to the Gradual Global internationalization process. The statistical

significance of a coefficient indicates the extent to which the corresponding explanatory variable

affects the marginal utility of the Born Global relative to the Gradual Global internationalization

process.

Result from Table 2 suggests that belonging to the ICT sector has a negative effect on a

firm’s choice on its internationalization process. This result seems puzzling, since numerous

studies (for example, De La Torre & Moxon, 2001; Dunning & Wymbs, 2001) have documented

the role of ICT in promoting the emergence of the Born Global phenomenon. However, a study

by Baldwin and Gu (2003) indicated that due to slower technological progress within the

high-technology sector in Canada compared with that in the US between 1996 and 2003,

productivity growth declined among Canadian manufacturers but increased remarkably among

5 In unreported regressions, I included both of the pobit model and the logit model, and conducted a normality test.

The results of the logit model are very similar to those of the probit model.


their US counterparts. Therefore, the negative impact of ICT sector on a Canadian manufacture’s

choice on the Born Global strategy is consistent with the findings by Baldwin and Gu.

Table 2. Probit Regression Results: What becomes of Born Global?

(Dependent Variables: BG=1 if a firm is classified as a Born Global and BG=0 otherwise.) Variables Coefficient Std. Err. Initial Size

= )log( 1employees

-0.2812*** 0.0311


= )ker/log( 1worrevenue

-0.2644*** 0.0654

Initial Wage Rates

= )ker/log( 1worpayroll

-0.1318 0.0900

ICT -0.3041* 0.1720

Foreign-owned 0.2034 0.2851

Cohort (reference: 1997) 1998 0.1051 0.1117

1999 0.3186*** 0.1112

2000 0.3347*** 0.1200

2001 0.4426*** 0.1168

2002 0.6911*** 0.1298

2003 0.9951*** 0.1505

2004 1.0795*** 0.2078

Sector dummies yes Province dummies yes Observations 1959 Log likelihood -990.09171 LR chi2 231.20 Pseudo R2 0.1045


Predicted Probability 0.2537 (0.4352)

Observed Probability 0.2557 (0.1506)

Note:

a. N = 1959. b. The complete results of Table 3 are reported in Table D in Appendix D. c. Value of standard errors is reported in parentheses. d. ***, ** and * indicate statistical significant at the 1%, 5% and 10% levels, respectively.


Result in Table 3 suggests that, everything else being equal, the smaller (in terms of number

of employees) the firm at the founding of the company, the more likely it will choose the Born

Global internationalization process. This result is consistent with Cavusgil et al. (2008, p15)’s

argument that smaller firms are more adaptable and have quicker response times to new ideas

and technologies. Consequently, smaller firms are more likely to export intensively at the

founding of the company.

Preferably, I would choose a measure of total factor productivity (TFP) to compare the

efficiency of Born Global and Gradual Global firms. Unfortunately, the available data does not

permit construction of a TFP index. Hence, I used the apparent labor productivity (APT),

revenue per worker, as a proxy for firms’ efficiency. Result from Table 3 suggests that

everything else being equal, the less productive the firm at the founding of the company, the

more likely it will choose the Born Global internationalization process.

Furthermore, the results in Table 3 suggest that, everything else being equal, the later the

firm starts its business, the more likely it will choose the Born Global internationalization

process. This finding is consistent with previous studies that use macro factors, such as shrinking

transportation and communication costs (Holstein, 1992) and “improvements in global

telecommunications and transport networks, combined with increasingly liberalized global

trading regimes” (Fan and Phan, 2007, p1113) as an explanation of the emergence of the Born


Global phenomenon. This finding may also suggest that Born Global has become a popular

internationalization process among small and medium-sized manufactures in Canada.

According to the estimated coefficients of the probit model, the mean value of Canadian

newly established small and medium-sized manufactures’ predicted probability of adopting the

Born Global internationalization process is 25.57% while the mean value of the observed

probability is 25.37%. Figure 2 plots the observed and predicted probability for a new firm to

choose the Born Global internationalization process in relation to the year it starts its business.

An inspection in Figure 2 shows two interesting results. First, the value of predicted probability

is very close to the value of observed probability for a firm to adopt the Born Global

internationalization process. This may suggest that my model of internationalization process

choice fits well and has a high power of prediction. Second, the later a firm starts its business,

the more likely it will choose the Born Global internationalization process.

Figure 2. The observed and predicted probability for a new firm to choose the Born Global internationalization process

Predicted Probability

Observed Probability

.15

.2

.25

.3

.35

.4

.45

.5

BG

%

1997 1998 1999 2000 2001 2002 2003 2004year start business

5.3 The Impact of Born Global Strategy on the Exit of New Exporters

Plotted in the figure 3, the survival rate at each moment indicates the probability that a new

exporter will keep exporting at any time until that moment. An inspection in Figure 3 shows that


the probability of survival in the export market of Gradual Global firms is slightly higher than

that of Born Global firms although the initial difference decreases over time.

The perception that Born Global firms have a lower survivability in the export market, as

shown in Table 3, is based on the fact that Born Global firms were smaller and less productive at

the founding of the company than Gradual Global firms. As such, the lower survivability in the

export market of Born Global firms could be a result of their poor initial conditions. A recent

study by Zettinig and Benson-Rea (2008) proposed a co-evolutionary approach to explain the

mechanisms that might secure the long-run survival and growth of Born Globals. The authors

argued that initial conditions of a firm, such as its ability to explore new knowledge, are essential

to its long-run survival and growth. Therefore, it is critically important to control for the

founding conditions of firms to assess the ceteris paribus effect of the Born Global

internationalization process on the survivability of firms.

Figure 3. Exit of New Exporters

Gradual Globals

Born Globals

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 1 2 3 4 5 6 7 8 9analysis time

Kaplan-Meier survival estimates

The key results of estimating several specifications of the hazard model are reported in Table

4. The details of the estimated results are reported in Table D in Appendix D. The coefficients in

Table 4 show the effect of the explanatory variables on the survivability of firms. A positive


coefficient indicates increased risk and a negative coefficient indicates a reduced risk. The

statistical significance of a coefficient indicates the extent to which the corresponding

explanatory variable affects the survivability of firms.

Table 4. The Impact of Born Global on the Export Market Survivability of New Exporters: Regression Results from the Cox Proportional Hazard Model

All Firms All Firms All Firms (2-stage)

All Firms (SSIV)

1 2 3 4 BG 0.0578*** -0.1582 -0.0315 (0.0222) (0.1319) (0.0604 ICT -0.0228 -0.0202 -0.0085 -0.0733 (0.0476) (0.0475) (0.0491) (0.0721 Foreign-owned -0.0165 -0.0135 -0.0067 -0.0273 (0.0351) (0.0350) (0.0359) (0.0629 US first -0.0113 -0.0120*** -0.0113 -0.0015 (0.0233) (0.0019) (0.0233) (0.0324 Initial Size = )log( 1employees

-0.0287** -0.0331*** -0.0418*** -0.0192 (0.0116) (0.0118) (0.0163) (0.0252)

Initial Labor Productivity = )ker/log( 1worrevenue

-0.0119*** -0.0144*** -0.0301*** -0.0321*** (0.0043) (0.0007) (0.0036) (0.0028)

Initial Wage Rates = )ker/log( 1worpayroll

0.0547* 0.0551* 0.0617* 0.0710 (0.0316) (0.0315) (0.0324) (0.0490)

Size = )log( temployees

-0.0026 -0.0027*** -0.0024*** -0.0058 (0.0025) (0.0007) (0.0003) (0.0040)

Labor Productivity = )ker/log( tworrevenue

-0.0125*** -0.0116*** -0.0125*** -0.0046** (0.0044) (0.0044) (0.0043) (0.0021)

Wage Rates = )ker/log( tworpayroll

0.0207*** 0.0200** 0.0206*** 0.0126 (0.0072) (0.0070) (0.0072) (0.0098)

Products = )log( tproducts

-0.0044** -0.0035*** -0.0041* -0.0057* (0.0023) (0.0006) (0.0023) (0.0034)

Destinations = )log( tnsdestinatio

-0.0039** -0.0036* -0.0039* -0.0019* (0.0020) (0.0020) (0.0020) (0.0009)

Re-enter

-0.2489*** -0.2471*** -0.2303*** -0.0955 (0.066)1 (0.0679) (0.0685) (0.0718)

Sector dummies yes yes yes yes


Province dummies yes yes yes yes Cohort dummies yes yes yes yes Wald chi2 1170.48 1288.92 1353.92 202.02 Observation 7287 7287 7287 3605 Firms 2072 2072 2072 958

Log pseudo likelihood

-34169.979 -34168.423 -34167.766 -14881.442

Note:

a. The complete results of Table 4 are reported in Table D in Appendix D. b. Robust standard errors (corrected for clustering at the firm level) are in parentheses. c. ***, ** and * indicate statistical significant at the 1%, 5% and 10% levels, respectively.


In order to demonstrate the importance of taking into account the internationalization

process on empirical estimates on the survivability of firms in the export market, I reported the

results of a regressions with and without iBG in column 2 and column 1, respectively. To

demonstrate the importance of taking into account the endogeneity of empirical estimates, I

reported the results of a regression with iBG^

, the estimated probability for firm i to choose

the Born Global internationalization process that is estimated from Table 3 in column 3. The

regression results with the Split-Sample Instrumental Variable (SSIV) methodology is reported

in column 4.

Table 4 shows the pseudo-likelihood ratio test statistics of models with different

specifications. Improvement is recorded if a firm’s internationalization process is included in the

model, since the log-pseudo-likelihood statistics increase from -34169.979 to -34168.423 as we

move from column 1 to column 2. Meanwhile, the estimated coefficients do not change much.

Further improvement is recorded when the selectivity bias is corrected, since the

log-pseudo-likelihood statistics increase from -34168.423 to -34167.766 as we moved from

column 2 to column 3.


At the top of Table 4, the coefficient 0.0578 of variable BG (column 2) indicates that Born

Global firms experience a hazard rate which is about 5.95% (i.e., 10578.0 −e ) higher than that of

Gradual Global firms, controlling for a wide range of factors that might affect the survivability of

firms in the export market. Such evidence supports Hypothesis 1, that is, everything else being

equal, Born Globals have a higher probability of exit from exporting than Gradual Globals.

When the estimates are corrected for the selectivity bias, shown in column 3, the magnitude

of this result changes, and the statistical significance of this result vanishes. I examined the

robustness of my results by applying the SSIV estimation. This result is consistent as the SSIV is

applied, shown in column 4. As such, the Born Global internationalization process no longer has

a negative impact on the survivability of firms in the export market. This result supports

Hypothesis 2, and suggests that estimations from models that do not account for endogeneity of

internationalization strategy choice may lead to incorrect conclusions.

5.4 Exit from Exporting of Born Global and Gradual Global Firms

In order to investigate the determinants on the survivability of Born Global and Gradual

Global firms in the export market, I reported the results of separate regressions for Born Global

and Gradual Global firms in Table 5 column 5 and column 6, respectively. It appears from Table

5 that Born Global and Gradual Global firms have rather different determinants of exit from

exporting. Shown in Table 5 column 5, for Born Globals, a decrease in the initial and current

wage rates of the firm, and a decrease in the variety of products exported of the firm, decrease its

hazard rate of exit, and thus, increase its survivability in the export market. Shown in Table 5

column 6, for Gradual Globals, a decrease in the initial size and current labor productivity of the

firm, an increase in the variety of products exported, an increase in the export destinations, and

choosing the US as the firm’s first export destination, decrease its hazard rate of exit, and thus,

increase its survivability in the export market.

Table 5. The Determinants of Exit from Exporting: Born Globals and Gradual Globals


Born Global Gradual Global 5 6 1 ICT 0.0590 -0.0609 (0.0867) (0.0551) 2 Foreign-owned -0.0679 -0.0056 (0.0646) (0.0457) 3 US first -0.0177 -0.0021** (0.0462) (0.0010) 4 Initial Size

= )log( 1employees -0.0210 -0.0469***

(0.0194) (0.0150) 5 Initial Labor Productivity

= )ker/log( 1worrevenue -0.0424 -0.0095

(0.0238) (0.0220) 6 Initial Wage Rates

= )ker/log( 1worpayroll 0.0749*** 0.0533

(0.0021) (0.0392) 7 Size

= )log( temployees

-0.0044 -0.0018

(0.0257) (0.0033)

8 Labor Productivity = )ker/log( tworrevenue

-0.0087 -0.0122** (0.0066) (0.0057) 9 Wage Rates

= )ker/log( tworpayroll

0.0202** 0.0128

(0.0094) (0.0101)

10 Products

= )log( tproducts

0.0071* -0.0023***

(0.0036) (0.0007)

11 Destinations

= )log( tnsdestinatio

-0.0034 -0.0074***

(0.0035) (0.0029)

12 Re-enter

-0.5480** -0.1924*** (0.2526) (0.0538) 13 Sector dummies yes yes 14 Province dummies yes yes 15 Cohort dummies yes yes 16 Wald chi2 86.88 913.71 17 Observation 2125 5162 18 Firms 521 1551

19 Log pseudo likelihood -8290.7449 -22829.628 Note:


a. The complete results of Table 5 are reported in Table D in Appendix D. b. Robust standard errors (corrected for clustering at the firm level) are in parentheses. c. ***, ** and * indicate statistical significant at the 1%, 5% and 10% levels, respectively.


I found that current labor productivity has a negative impact on the hazard rate of exit, and

thus, a positive impact on the survivability of Gradual Global firms in the export market. This

result is consistent with the predictions from the Melitz (2003) model. However, neither initial

nor current labor productivity has a statically significant impact on the survivability of Born

Global firms. Therefore, Hypothesis 2a that more productive firms are more likely to survive in

the export market is partially supported.

Neither initial nor current wage rates have a statically significant impact on the survivability

of Gradual Global firms in the export market. Moreover, counter to expectations, both initial and

current wage rates have a positive impact on the hazard rate of exiting for Born Global firms.

Therefore, Hypothesis 2b that firms pay higher wage are more likely to survive in the export

market is not supported. This result seems puzzling because one would expect that firms that are

able to survive in the export market are usually good firms that are able to pay higher wages to

their employees. However, these results are consistent with the predictions from Albornoz et al.

(2009) if we consider wage as a proxy for labor costs instead of labor productivity.

Specifically, building on the original Melitz model, Albornoz et al. (2009) developed a

mechanism that export profitability is uncertain for a firm before it starts exporting. Everything

else being the same, both lower labor costs and higher labor productivity will lead to a higher

level of profitability. Firms with lower labor costs may anticipate a higher level of profitability

before they start exporting. Such firms will export intensively at the founding of the company

and become Born Globals. Moreover, such firms are able to continuously exporting if they could

maintain their advantage in lower labor costs. On the other hand, firms learn about their

productivity only after they start exporting. If firms export just to learn about their profit


potential, they will export only the minimum necessary for effective learning and become

Graduals Global. Moreover, such firms are able to continuously exporting if they could maintain

their advantage in high labor productivity.

For Gradual Global firms, both export product diversification and export market

diversification have a negative impact on the hazard rate of exit, and thus, a positive impact on

the probability of survival in the export market. These results are consistent with the results from

Sabuhoro et al. (2006). For Gradual Global firms, however, their export market survivability

does not seem to be affected by their export market diversification. Moreover, it appears that

export product diversification has a positive impact on the hazard rate of exit for Born Global

firms. Therefore, Hypothesis 3a that firms exporting to more destinations are more likely to

survive in the export market is partially supported. Similarly, Hypothesis 3b that more

product-diversified firms are more likely to survive in the export market is partially supported.

Nevertheless, my findings are consistent with a previous study by Chetty and Campbell-Hunt

(2004) on New Zealand firms. Chetty and Campbell-Hunt showed that Born Global firms are

characterized by having a narrow product range and serving a “specialized product for niche

markets”; regional firms, on the other hand, have “diversified product range to broad market

segments”. Therefore, export product and market diversification may not have as much of a

positive impact on the export market survivability for Born Global firms as for Gradual Global

firms.

Choosing the “easiest” and “closest” market (the US) as the first export destination has a

statically significant negative impact on the hazard of exit of Gradual Globa but no impact on the

hazard of exit of Born Global firms. Hence, both Hypothesis 4a and Hypothesis 4b are

supported.

The effect of foreign ownership on firms’ survivability in the export market will depend

mainly on whether the foreign investors seek to establish an exporting platform or to find a new


market (Dunning, 1977). Additionally, previous empirical evidence on the economic

performance of foreign and domestically owned Canadian establishments (Globerman et al.,

1994) showed that foreign-owned Canadian establishments have significantly higher value added

per worker and pay higher wages than domestic-owned Canadian establishments. However, such

difference disappeared once the authors controlled for factors such as size and capital intensity of

the firms. Therefore, it is not surprising to find that foreign ownership has a negative but

insignificant impact on the hazard rate of exit for both Born Global and Gradual Global firms.

Finally, my results show that previous export experience has a statistically significant

negative impact on the hazard of exit from exporting, and hence, a positive impact on the

survival of exporters, regardless of their internationalization process.

I was concerned with the possibility that these results might have been decisively affected by

the alternative numerical definition of Born Global firms. Accordingly, I ran the same

regressions as above using Knight and Cavusgil (1996)’s definition (three years and 25%

thresholds) on Born Globals. Those results are qualitatively identical to the results reported in

Table 4 and Table 5.

6 Conclusions

This study seeks to answer the following questions: What is the impact of Born Global, the

accelerated internationalization process, on the survivability of firms in the export market? What

are the determinants of survival of Born Global and Gradual Global firms in the export market?

To provide answers to above questions, I analyzed the export market survival patterns of

1,959 newly established small and medium-sized Canadian manufacturers between 1997 and

2005. A firm is classified as a Born Global if it begins to export within two years of its inception

and exports no less than 25% of its total revenue during the first year of its export activity;

otherwise, a firm is classified as a Gradual Global. The percentage of firms that are classified as

Born Global is 25.37%.


After controlling for a wide range of factors that might affect the survivability of firms in the

export market, it appears that Born Global has a significant negative impact on the survivability

of firms in the export market. After accounting for the self-selection associated with a firm's

choice on its internationalization process, however, Born Global no longer affects firm s’

survivability in the export market. These results suggest that Born Global is a strategic choice for

a particular group of young firms that appear in my data to be relatively small and less

productive at the founding of the company. Such firms would have a less chance of survival in

the export market if they have chosen the traditional Gradual Global internationalization process.

My findings suggest that Born Global and Gradual Global firms have very different

determinants of exit from exporting. The factors that determine the export market survivability of

Gradual Globals firms are consistent with previous empirical studies (Harris and Li, 2007;

Sabuhoro et al., 2006), that firms are larger, more productive, export more products, export to

more markets, and choosing the US as the first export destination have a better chance of

survival in the export market.

Unlike Gradual Global firms, Born Global firms will have a higher survivability in the

export market with a decrease in both the variety of products they export and the wages they pay.

These results could have important implications for the policy-makers in improving the design

and targeting of Canada's trade and investment promotion programs. These results also have

important implications for researchers in understanding that internationalization is a mixed bag

containing different types of firms, who have different advantages and disadvantages in

competing in the global export market.

The absence of quality longitudinal data on the international activity of businesses has

prevented researchers’ ability to understand the underlying mechanisms for real world business

phenomena and how firms respond to challenges in the new era of globalization. It has also

prevented the government’s ability to propose comprehensive and effective policies to promote


trade and foreign direct investment. Guided on the theoretical framework in the literature, my

on-going research is focused on empirical investigation on the complex and dynamic nature of

the internationalization process of firms. More specifically, I will attempt to investigate the

evolution and growth of Born Global and Gradual Global firms before and after their

participation in the global market.


REFERENCES Albornoz, F., Calvo, H., Corcos, G., and Ornelas, E. (2009), “Sequential Exporting”, working in

progress. Angrist, Joshua D. and Krueger, Alan B. (1995),“Split-Sample Instrumental Variables Estimates

of the Returns to Schooling.” Journal of Business and Economic Statistics, 13:2, 225–35. Audretsch, D.B. and Mahmood, T. (1995), “New firm survival: new results using a hazard

function”, Review of Economics and Statistics 77(1): 97–103. Baldwin, J.R. and Gorecki, P.K. (1991), “Productivity Growth and the Competitive Process: The

Role of Firm and Plant Turnover”, In P.A. Geroski and J. Schwalbach. Entry and Market Contestability: An International Comparison. Oxford: Basil Blackwell. 244-56.

Baldwin, J., & Gu W. (2003), “Participation in export markets and productivity performance in Canadian manufacturing”, Research Paper No. 011, Statistics Canada Publication, Ottawa.

Bell, J. (1995), “The Internationalization of Small Computer Software Firms”, European Journal of Marketing, 29(8), 60-75.

Bernard, A.B., and J. Bradford Jensen (1999), “Exceptional exporter performance: cause, effect, or both?”. Journal of International Economics 47(1), p1-25.

Boter, H., & Holmquist, C. (1996), “Industry characteristics and internationalization processes in small firms”. Journal of Business Venturing, 11: 471-487.

Brush, C. G. (1992), “Factors Motivating Small Companies to Internationalize: the Effect of Firm Age”, Unpublished dissertation, Boston University, Boston.

Cavusgil, S. T., Knight, G., Riesenberger, J. R. (20080, International Business: Strategy, Management, and the New Realities, ISBN-10: 0131738607.

Chetty S. & Campbell-Hunt, C. (2004), “A strategic approach to internationalisation: A traditional versus a 'born-global' approach”. Journal of International Marketing, Vol. 12 No.1, pp.57-81.

Clerides, S. S. Lach, J. Tybout (1998), “Is learning by exporting important? Micro-dynamic evidence from Colombia, Mexico, and Morocoo”, Quarterly Journal of Economics 113(3), p903-947.

Coviello, N. E. and Jones, M. V. (2004), “Methodological issues in international entrepreneurship research”. Journal of Business Venturing, 19(4), 485–508.

Cox, D.R. (1972), Regression Models and Life Tables, Journal of the Royal Statistical Society, 34, 187-220.

De La Torre, J., & Moxon, R. W. (2001), “E-commerce and global business: The impact of the information and communication technology revolution on the conduct of international business”. Journal of International Business Studies, 32(4): 12–25.

Delgado, M.A. J.C. Farinas, S. Ruano (2002), “Firm productivity and export markets: a non-parametric approach”, Journal of International Economics 57, p397-422.


Dunning, J. H. (1977), “Trade, location and economic activity and the multinational enterprise: A search for an eclectic approach”. InB. Ohlin, P. Hesselborn, & P. Wijkman (Eds.), The international allocation of economic activity. London: MacMillan.

Dunning, J.,&Wymbs, C. (2001), “The challenge of electronic markets for international business theory”, International Journal of the Economics of Business, 8(2): 123–145.

Eaton, J., S. Kortum and F. Kramarz (2005), “An anatomy of International Trade, Evidence from French firms,” mimeo University of Minnesota and NYU.

Eaton, J., Eslava, M., Kugler, M., and Tybout, J. (2007), “Export Dynamics in Colombia: Firm-Level Evidence”, NBER Working Paper Series, working paper 13531.

Fan, T., and Phan, P. (2007), “International new ventures: revisiting the influences behind the ‘born-global’ firm’, Journal of International Business Studies, 38: 1113-1131.

Forsgren, M. and Johanson, J. (1975), Internationall företagsekonomi. Stockholm: Norstedts. Freeman J, Carroll GR, and Hannan MT (1983), “The liability of newness, age dependence and

organizational death rates”, American Sociological Review, Vol. 48 pp.692-710. Gabrielsson et al. (2008), “Conceptualizations to Advance Born Global Definition: A Research

Note”, Global Business Review, Vol. 9, No. 1, 45-50. Geroski, P. A. (1995), “What Do We Know about Entry?”, International Journal of Industrial

Organization, 13, 421–440. Globerman, Steven, Ries, John C., and Vertinsky, Ilan (1994), “The economic performance of

foreign affiliates in Canada”, Canadian Journal of Economics, Vol. XXVII, No. 1, pp. 143-156.

Johanson, J., and Wiedersheim-Paul, F. (1975), “The Internationalization Process of the Firm-Four Swedish Case Studies”, Journal of Management Studies, Vol. 12, pp. 305-322.

Johanson, J., and Vahlne, J.E. (1990), ”The mechanism of internationalization”, International Marketing Review, Vol. 7, No. 4, pp. 11 -24.

Harris R and Q., Li (2007), “Determinants of exit from exporting”, report to UKTI. Holstein, W. J. (1992), “Little companies, big exports”. Business Week,13(April): 70–72. Isgut, A.E. (2001), “What’s different about exporters? Evidence from Colombia manufacturing”,

Journal of Development Economics 37(5), p 57-82. Keeble, D., Lawson, C., Smith, H. L., Moore, B., & Wilkinson, F. (1998), “Internationalization

processes, networking and local embeddedness in technology-intensive small firms”. Small Business Economics, 11: 327-342.

Keupp, M. M., and Gassmann, O. (2009), “The past and the future of international entrepreneurship: A review and suggestions for developing the field”, Journal of Management, 35(Annual Review Issue 2009), 600-633.

Knight, G.A. (1997), “Emerging paradigm for international marketing: the Born Global firm”, PhD thesis.

Knight, G. A., and Cavusgil, S. T. (1996), “The born global firm: A challenge to traditional internationalization theory”, Advances in International Marketing, 8: 11–26.


Krugman, P.R. (1979), “Increasing returns, monopolistic competition, and international Trade”, Journal of International Economics 9, 469-479.

Lam W. L. and White, P. L. (1999), “An adaptive choice model of the internationalization process”, The International Journal of Organizational Analysis, Vol. 7, No. 2, April, pp.105-34.

Lancaster, T., (1990), The econometric analysis of transition data. Cambridge University Press, Cambridge.

Mata, J., P. Portugal and P. Guimaraes (1995), “The Survival of New Plants, Start-up Conditions and Post-Entry Evolution”, International Journal of Industrial Organization, 13(4), 459–481.

Masten, S.E. (1993), “Transaction costs, mistakes and performance: assessing the importance of governance”, Managerial and Decision Economics 14(2): 119–129.

Madsen, T. K., Servais, P., (1997), “The Internationalization of Born Globals: an Evolutionary Process?”, International Business Review, Vol. 6, pp. 561-583.

Madsen, T. K., Rasmussen, E. S., and Servais, P. (2000), “Differences and similarities between born globals and types of exporters”, Advances in International Marketing, 10: 247–265.

Mayer, Thierry, and Ottaviano, Gianmarco I. P., (2008), “The Happy Few: The Internationalisation of European Firms--New Facts based on Firm-level Evidence”, Intereconomics, Volume 43, Number 3 / May, 135-148.

McDougall, Patricia P. and Oviatt, Benjamin M. (2000), “International Entrepreneurship: The Intersection of Two Research Paths,” Academy of Management Journal, 43 (5), 902–906.

McKinsey and Co. (1993), “Emerging Exporters: Australia’s High Value-Added Manufacturing Exporters”. Melbourne: McKinseyand Co., Australian Manufacturing Council.

Meckl, R., and Schramm, R., (2005), "Empirical evidence for a theory of international new ventures", Jenaer Schriften zur Wirtschaftswissenschaft, Nr. 06.

Melitz, M.J. (2003), “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity”. Econometrica, 71(6), 1695-1725.

Moen, Øystein and Per Servais (2002), “Born Global or Gradual Global? Examining the Export Behavior of Small and Medium-Sized Enterprises,” Journal of International Marketing, 10 (3), 49–72.

Mudambi, R.,and Zahra, S.A., (2007), “The survival of international new ventures”, Journal of International Business Studies, 38, 333–352

Oviatt, B. M., and McDougall, P. P. (1994), “Toward a Theory of International New Ventures”, Journal of International Business Studies 24, 45-64.

Oviatt, B. M., & McDougall, P. P. (1997), “Challenges for internationalization process theory: The case of international new ventures”, Management International Review, 37: 85-99.

Rennie, M. W. (1993), “Born Global”, McKinsey Quarterly 4, 45-52. Sabuhoro J. B, Larue, B., Gervais, Y., (2006), “Factors determining the success or failure of

Canadian establishments on foreign markets: a survival analysis approach”, The Internationals Trade Journal, Volume XX, No. 1, p33-73.


Sapienza, H. J., Autio, E., George, G., Zahra, S. A., (2006), “A Capabilities Perspective on the. Effects of Early Internationalization on Firm Survival and Growth”, Academy of Management Review, Vol. 31, Issue 4, p 914-933.

Shrader, R. C. (2001), “Collaboration and performance in foreign markets: The case of young high-technology manufacturing firms”, Academy of Management Journal, 44: 45-60.

Shaver, M. (1998), “Accounting for endogeneity when assessing strategy performance: Does entry mode choice affect”. Management Science, 44 (4): 571-585.

Stinchcome, A.L. (1965), “Social Structure and Organizations”, in J. March (ed.) Handbook of Organizations, Rand McNally: Chicago, IL, pp: 142–193.

Ursic, M.L., Czinkota, M.R. ( 1984), "An experience curve explanation of export expansion", Journal of Business Research, Vol. 12 pp.159-68.

Wagner, Joachim (2003), "Are Nascent Entrepreneurs Jacks-of-All-Trades? A Test of Lazear's Theory of Entrepreneurship with German Data," IZA Discussion Papers 911, Institute for the Study of Labor (IZA).

Welch, L. S. and Luostarinen, R. (1988), “Internationalisation: evolution of a concept”, Journal of General Management, Vol.34, Winter, pp.34-57.

Zahra, S.A., Ireland, R.D. and Hitt, M.A. (2000), “International expansion by new venture firms: international diversity, mode of market entry, technological learning, and performance”, Academy of Management Journal, Vol. 43 No. 5, pp. 925-50.

Zahra, S. A. (2005), “A theory of international new ventures: a decade of research”, Journal of International Business Studies, Volume 36, Number 1, January 2005 , pp. 20-28(9).

Zettinig, Peter and Benson-Rea, M. (2008), “What becomes of International New Ventures? A coevolutionary approach”, European Management Journal, Vol. 26(6), p. 354-365.


Appendix A: Variables of Interest

Table A1. Variable definitions, Descriptive statistics, and Sources

Name Definition Mean S.D. Max Min Source BEBY the first year a firm appears

in the BR database 1999.43 1.95 1997 2004 BR

ERBY the first year a firm appears in the ER database

2000.75 2.00 1997 2004 ER

ERDY the last year a firm appears in the BR database

2004.35 1.28 1997 2005 ER

Duration Spell duration in years 3.38 2.01 1 9 ER Censor =1 if the firm exports in

2005 0.71 0.21 0 1 ER

Re-enter =1 if the firm re-enters the export market

0.3 0.16 0 1 ER

ExAge =ERBY-BRBY+1, the firm’s export start-up age

1.32 1.59 0 7 ER, BR

Exports Annual value of exports, in millions of Canadian dollars

0.70 2.17 0.002 72 ER

Revenue Annual value of revenue, in millions of Canadian dollars

1.99 3.95 0.03 75 BR

EI Exports/Revenue, export intensity

0.36 0.34 0 1 ER, BR

BG =1 if the firm is classified as a Born Global and 0 otherwise

0.25 0.44 0 1 ER, BR

ICT =1 if the firm belongs to the information and communication sector and 0 otherwise

0.30 0.46 0 1 BR

Foreign =1 if the firm is owned by a foreign country

0.02 0.13 0 1 BR

US first =1 if the US is the firm’s first export destination

0.82 0.32 0 1 ER

Employees number of employees the firm hired

21.34 32.69 1 591 LEAP

Wage Rates

Payroll/Employees, in thousands of Canadian

29.02 15.81 0.9 247 LEAP


dollars Labor Productivity

Revenue/Worker, in thousands of Canadian dollars

98.76 94.42 0 3976 BR, LEAP

Destinations number of export destinations

1.65 2.43 1 54 ER

Products number of exported products 3.44 4.09 1 80 ER Sector =1 if the firm belongs to that

sector and 0 otherwise . . 0 1 ER

Province =1 if the firm is located in that province and 0 otherwise

. . 0 1 ER

Cohort =1 if the firm starts its business in that year and 0 otherwise

. . 0 1 BR

Year =1 if the firm starts to export in that year and 0 otherwise

. . 0 1 ER



Appendix B: Characteristics of firms that are grouped by export start-up age and initial export intensity

In this study, I use export start-up age and initial export intensity to classify Born Global.

Before adopting any numerical definition of Born Global that has been used by previous studies,

this section explores the characteristics of firms that are grouped by different thresholds of these

two variables grouped in a 5*5 matrix.

Table B Matrix of firms that are decomposed by export start-up age and initial export intensity

B1 Number of Firms Initial export intensity (%) 50-100 25-50 12.5-25 6.25-12.5 0-6.25 total

Export start-up age

0 197 169 132 118 200 816 1 58 73 94 74 173 472 2 37 36 52 50 116 291 3 14 20 21 25 73 153 4+ 18 32 37 30 110 227

total 324 330 336 297 672 1,959

B2 )log( 1revenue Initial export intensity (%) 50-100 25-50 12.5-25 6.25-12.5 0-6.25 total

Export start-up age

0 0.54 0.96 1.00 1.52 1.86 1.17 1 0.45 0.75 0.57 0.63 1.84 1.06 2 0.59 0.51 1.44 0.43 1.19 0.95 3 0.15 0.43 0.61 0.28 1.56 0.94 4+ 0.42 1.19 0.33 0.71 1.72 1.18

total 0.50 0.86 0.85 0.93 1.68 1.09

B3 )log( 1employees Initial export intensity (%) 50-100 25-50 12.5-25 6.25-12.5 0-6.25 total

Export start-up age

0 1.76 1.98 2.17 2.26 2.59 2.15 1 1.69 2.06 1.99 2.38 2.71 2.29 2 1.73 2.25 1.86 2.21 2.69 2.28 3 1.59 2.03 2.14 2.34 2.72 2.38 4+ 1.60 2.53 1.98 2.29 3.02 2.57

total 1.73 2.08 2.05 2.29 2.72 2.27


B4 )ker/log( 1worpayroll Initial export intensity (%) 50-100 25-50 12.5-25 6.25-12.5 0-6.25 total

Export start-up age

0 3.02 3.15 3.08 3.15 3.21 3.12 1 3.11 3.14 3.18 3.19 3.31 3.21 2 3.11 3.11 3.11 3.17 3.21 3.16 3 2.93 3.33 3.26 3.19 3.38 3.28 4+ 3.17 3.08 2.99 3.18 3.34 3.21

total 3.05 3.14 3.11 3.17 3.27 3.17

B5 )ker/log( 1worrevenue Initial export intensity (%) 50-100 25-50 12.5-25 6.25-12.5 0-6.25 total

Export start-up age

0 10.95 11.06 11.00 11.19 11.32 11.11 1 11.03 11.04 11.13 11.11 11.24 11.14 2 11.01 10.89 10.92 10.98 11.06 10.99 3 10.81 10.95 11.17 10.89 11.25 11.10 4+ 11.06 11.03 10.95 11.08 11.25 11.13

total 10.97 11.03 11.03 11.10 11.24 11.10

B6 )log(exp 1orts Initial export intensity (%) 50-100 25-50 12.5-25 6.25-12.5 0-6.25 total

Export start-up age

0 -1.46 -2.01 -2.58 -3.10 -4.02 -2.62 1 -1.53 -1.89 -2.50 -2.85 -3.88 -2.85 2 -1.36 -1.62 -2.67 -3.03 -3.92 -2.93 3 -2.00 -1.58 -2.32 -2.80 -3.44 -2.80 4+ -1.73 -1.32 -2.63 -2.64 -3.25 -2.68

total -1.50 -1.85 -2.56 -2.95 -3.78 -2.74

B7 )log( 1nsdestinatio Initial export intensity (%) 50-100 25-50 12.5-25 6.25-12.5 0-6.25 total

Export start-up age

0 0.16 0.10 0.10 0.04 0.03 0.09 1 0.14 0.10 0.09 0.08 0.07 0.09 2 0.22 0.08 0.13 0.19 0.09 0.13 3 0.10 0.18 0.12 0.00 0.05 0.07 4+ 0.19 0.10 0.21 0.12 0.10 0.13

total 0.16 0.10 0.12 0.08 0.06 0.10


B8 )log( 1products Initial export intensity (%) 50-100 25-50 12.5-25 6.25-12.5 0-6.25 total

Export start-up age

0 0.89 0.72 0.61 0.53 0.31 0.61 1 0.77 0.70 0.68 0.53 0.50 0.61 2 0.80 0.79 0.56 0.50 0.42 0.55 3 0.83 0.71 0.88 0.43 0.41 0.55 4+ 0.79 0.92 0.49 0.62 0.43 0.56

total 0.85 0.74 0.62 0.52 0.41 0.59

B9 ICT Initial export intensity (%) 50-100 25-50 12.5-25 6.25-12.5 0-6.25 total

Export start-up age

0 8.12% 8.28% 7.58% 10.17% 5.50% 7.72% 1 3.45% 2.74% 7.45% 10.81% 6.36% 6.36% 2 13.51% 5.56% 15.38% 4.00% 7.76% 8.93% 3 7.14% 10.00% 4.76% 0.00% 13.70% 9.15% 4+ 11.11% 3.13% 2.70% 6.67% 7.27% 6.17%

total 8.02% 6.36% 8.04% 8.08% 7.29% 7.50%

B10 Foreign Initial export intensity (%) 50-100 25-50 12.5-25 6.25-12.5 0-6.25 total

Export start-up age

0 1.02% 1.78% 1.52% 0.00% 4.00% 1.84% 1 0.00% 1.37% 0.00% 1.35% 1.16% 0.85% 2 0.00% 0.00% 1.92% 0.00% 2.59% 1.37% 3 0.00% 5.00% 4.76% 0.00% 2.74% 2.61% 4+ 0.00% 3.13% 0.00% 0.00% 4.55% 2.64%

total 0.62% 1.82% 1.19% 0.34% 2.98% 1.68% Note: a. Variables ‘employees’, ‘payroll per worker’, and ‘revenue per worker’ are estimated in the first

year a firm started its business; variables ‘exports’, ‘products’, and ‘destinations’ are estimated in the first year a firm started to export.



Appendix C: Probit Regression Results

Table 2b: Estimating Born Global strategy choice

Dependent Variables: BG=1 if a firm is classified as a Born-Global and BG=0 otherwise. Variables Coef. Std. Err. Constant 3.2497*** 0.6457 ICT -0.3041* 0.1720 Foreign-owned 0.2034 0.2851 Initial Size = )log( 1employees

-0.2812*** 0.0311

Initial Labor Productivity = )ker/log( 1worrevenue

-0.2644*** 0.0654

Initial Wage Rates = )ker/log( 1worpayroll

-0.1318 0.0900

Sector(reference: Computer) Food -0.3507 0.2192 Beverage -0.2059 0.4251 Textile -0.4200 0.3398 Textile Product -0.6418* 0.3680 Clothing 0.0920 0.2164 Leather 0.2069 0.4980 Wood -0.3071 0.2013 Paper -0.3874 0.3737 Printing -0.6195** 0.2478 Petroleum 0.0416 0.7763 Chemical -0.2262 0.2211 Plastic & Rubber -0.1628 0.2049 Non-Metallic Mineral -0.4621* 0.2604 Metal 0.4820 0.3135 Fabricated Metal -0.3940* 0.1911 Machinery -0.2331 0.1758 Electronics -0.3046 0.2251 Transportation Equipments 0.0358 0.2203 Furniture -0.3737* 0.1998 Miscellaneous -0.3069 0.1929 Province (reference: Ontario) Newfoundland 0.3701 0.2541 Nova Scotia -0.0432 0.2349


New Brunswick -0.0643 0.0861 Quebec -0.5769** 0.2493 Manitoba -0.0168 0.3369 Saskatchewan 0.0708 0.1311 Alberta -0.1158 0.0969 British Columbia -0.010 0.092 Cohort (reference: 1997) 1998 0.1051 0.1117 1999 0.3186*** 0.1112 2000 0.3347*** 0.1200 2001 0.4426*** 0.1168 2002 0.6911*** 0.1298 2003 0.9951*** 0.1505 2004 1.0795*** 0.2078 Observations 1959 Log likelihood -990.09171 LR chi2 231.20 Pseudo R2 0.1045 Observed Probability 0.2537 (0.4352) Predicted Probability 0.2557 (0.1506)

Note:

a. The key results of Table 2b are reported in Table 2a. b. ***, ** and * indicate statistical significant at the 1%, 5% and 10% levels, respectively. Data Source: the Exporter Register, the Business Register, and the LEAP from Statistics Canada.


Appendix D: Cox proportional Hazard Regression Results

Table D: The Determinants of New Exporter Exit from Exporting

All Firms All Firms All Firms (FE)

All Firms (2-stage)

All Firms (SSIV)

Born Global Gradual Global

1 2 3 4 5 6 7 BG 0.0578*** 0.0615* -0.1582 -0.0315 (0.0222) (0.0326) (0.1319) (0.0604) ICT -0.0228 -0.0202 0.0222 -0.0085 -0.0733 0.0590 -0.0609 (0.0476) (0.0475) (0.0546) (0.0491) (0.0721) (0.0867) (0.0551) Foreign-owned -0.0165 -0.0135 0.0245 -0.0067 -0.0273 -0.0679 -0.0056 (0.0351) (0.0350) (0.1025) (0.0359) (0.0629) (0.0646) (0.0457) US first -0.0113 -0.0120*** -0.0284 -0.0113 -0.0015 -0.0177 -0.0021** (0.0233) (0.0019) (0.0417) (0.0233) (0.0324) (0.0462) (0.0010) Initial Size

= )log( 1employees

-0.0287** -0.0331*** -0.0335* -0.0418*** -0.0192 -0.0210 -0.0469*** (0.0116) (0.0118) (0.0192) (0.0163) (0.0252) (0.0194) (0.0150)


= )ker/log( 1worrevenue

-0.0119*** -0.0144*** -0.0177 -0.0301*** -0.0321*** -0.0424 -0.0095

(0.0043) (0.0007) (0.0286) (0.0036) (0.0028) (0.0238) (0.0220)

Initial Wage Rates

= )ker/log( 1worpayroll

0.0547* 0.0551* 0.0582 0.0617* 0.0710 0.0749*** 0.0533 (0.0316) (0.0315) (0.0463) (0.0324) (0.0490) (0.0021) (0.0392)

Size

= )log( temployees

-0.0026 -0.0027*** 0.0020 -0.0024*** -0.0058 -0.0044 -0.0018 (0.0025) (0.0007) (0.0056) (0.0003) (0.0040) (0.0257) (0.0033)


Labor Productivity = )ker/log( tworrevenue

-0.0125*** -0.0116*** -0.0099 -0.0125*** -0.0046** -0.0087 -0.0122** (0.0044) (0.0044) (0.0108) (0.0043) (0.0021) (0.0066) (0.0057)

Wage Rates

= )ker/log( tworpayroll

0.0207*** 0.0200** 0.0188 0.0206*** 0.0126 0.0202** 0.0128 (0.0072) (0.0070) (0.0151) (0.0072) (0.0098) (0.0094) (0.0101)

Products

= )log( tproducts

-0.0044** -0.0035*** 0.0026 -0.0041* -0.0057* 0.0071* -0.0023*** (0.0023) (0.0006) (0.0054) (0.0023) (0.0034) (0.0036) (0.0007)

Destinations

= )log( tnsdestinatio

-0.0039** -0.0036* 0.0058 -0.0039* -0.0019* -0.0034 -0.0074*** (0.0020) (0.0020) (0.0070) (0.0020) (0.0009) (0.0035) (0.0029)

Re-enter -0.2489*** -0.2471*** -0.2311* -0.2303*** -0.0955 -0.5480** -0.1924*** (0.066)1 (0.0679) (0.1184) (0.0685) (0.0718) (0.2526) (0.0538) Sector(reference: Computer) Food -0.0201 -0.0185 -0.0025 -0.0560 -0.0760 -0.0098 (0.0594) (0.0579) (0.0605) (0.0864) (0.1316) (0.0650) Beverage 0.0571 0.0574 0.0653 0.0529 0.0851 0.0452 (0.0557) (0.0560) (0.0560) (0.1459) (0.1017) (0.0675) Textile -0.3005** -0.2925** -0.2788** -0.4643* -1.1544** -0.1261 (0.1389) (0.1413) (0.1403) (0.2666) (0.6859) (0.1143) Textile Product -0.0409 -0.0329 -0.0133 -0.0516 0.0613 -0.0856 (0.0957) (0.0951) (0.0995) (0.1226) (0.1078) (0.1199) Clothing -0.1274* -0.1281* -0.1305 -0.1514 -0.0024 -0.2401** (0.0759) (0.0756) (0.0759) (0.1034) (0.1080) (0.1084) Leather 0.0799 0.0816 0.0694 0.0591 0.0585 0.1315 (0.0875) (0.0889) (0.0872) (0.1189) (0.1901) (0.1140)


Wood -0.0718 -0.0708 -0.0572 -0.1881** 0.0020 -0.1002 (0.0578) (0.0579) (0.0590) (0.0918) (0.1009) (0.0690) Paper 0.0077 0.0171 0.0282 -0.0554 0.0345 0.0242 (0.0793) (0.0800) (0.0809) (0.1183) (0.1734) (0.0940) Printing -0.1506** -0.1410* -0.1247 -0.2723** -0.0203 -0.1594 (0.0762) (0.0763) (0.0793) (0.1328) (0.1391) (0.0882) Petroleum 0.0828 0.0875 0.0834 0.0308 0.0474 0.0768 (0.0538) (0.0574) (0.0544) (0.0784) (0.1024) (0.0644) Chemical 0.0110 0.0150 0.0222 -0.0020 -0.0846 0.0556 (0.0539) (0.0547) (0.0546) (0.0763) (0.1247) (0.0579) Plastic & Rubber -0.0504 -0.0497 -0.0412 -0.0566 -0.0020 -0.0739 (0.0602) (0.0603) (0.0608) (0.0861) (0.1035) (0.0737) Non-Metallic Mineral 0.0139 0.0188 0.0381 -0.0494 0.1367 -0.0207 (0.0597) (0.0596) (0.0623) (0.0797) (0.1007) (0.0706) Metal -0.0519 -0.0638 -0.0818 -0.1400 -0.2612 0.0858 (0.0959) (0.0975) (0.0995) (0.1456) (0.2033) (0.0659) Fabricated Metal -0.0693 -0.0645 -0.0495 -0.0585 -0.0471 -0.0704 (0.0534) (0.0537) (0.0560) (0.0748) (0.1067) (0.0610) Machinery -0.0470 -0.0443 -0.0351 -0.1340* -0.0258 -0.0438 (0.0486) (0.0488) (0.0492) (0.0725) (0.0899) (0.0568) Electronics -0.0221 -0.0200 -0.0062 -0.1148 -0.0256 -0.0125 (0.0568) (0.0565) (0.0581) (0.0964) (0.1029) (0.0644) Transportation Equipments

-0.1016 -0.1062 -0.1050 -0.1881* -0.0377 -0.1336 (0.0676) (0.0678) 0.0676 (0.1048) (0.1214) (0.0843)

Furniture -0.1409** -0.1364** -0.1223 -0.2187** -0.1561 -0.1326* (0.0656) (0.0658) (0.0678) (0.0992) (0.1321) (0.0743)


Miscellaneous -0.0182 -0.0158 -0.0029 -0.0523 0.0485 -0.0486 (0.0537) (0.0539) (0.0542) (0.0762) (0.0929) (0.0658) Province (reference: Ontario) Newfoundland -0.1058 -0.0874 -0.0663 -0.0536 0.0506 . -0.0623 (0.1626) (0.1634) (0.2690) (0.1670) (0.1149) . (0.1623) Nova Scotia -0.0991 -0.1064 -0.0953 -0.1166 -0.1433 -0.0997 -0.0646 (0.1018) (0.1018) (0.1286) (0.1027 (0.1677) (0.1612) (0.1432) New Brunswick 0.0461 0.0469 0.0433 0.0461 0.0044 0.0848 0.0066 (0.0576) (0.0569) (0.0900) (0.0576 (0.1210) (0.0875) (0.0727) Quebec -0.0359 -0.0342 -0.0392 -0.0327 0.0109 -0.0345 -0.0194 (0.0285) (0.0286) (0.0384) (0.0287 (0.0424) (0.0511) (0.0337) Manitoba 0.0878** 0.0964** 0.1076 0.1098** 0.1011 0.2822*** 0.0844 (0.0427) (0.0431) (0.0957) (0.0466 (0.0813) (0.1070) (0.0552) Saskatchewan 0.0143 0.0244 0.0384 0.0180 0.0129 -0.1820 0.0461 (0.0872) (0.0884) (0.1496) (0.0869) (0.1123) (0.3017) (0.0878) Alberta -0.0138 -0.0154 -0.0025 -0.0177 0.0794 -0.0227 -0.0254 (0.0396) (0.0392) (0.0583) (0.0396) (0.0455) (0.0589) (0.0499) British Columbia -0.0095 -0.0039 -0.0009 -0.0036 0.0075 -0.0246 0.0164 (0.0289) (0.0290) (0.0428) (0.0294) (0.0472) (0.0571) (0.0338) Cohort (reference: 1997) 1998 0.0762 0.0803 0.0818 0.0726 0.0675 0.0239 0.1397 (0.0570) (0.0566) (0.0612) (0.0573) (0.0908) (0.0628) (0.0914) 1999 0.0826 0.0874 0.0892 0.0724 0.1797** -0.0527 0.2009** (0.0566) (0.0566) (0.0614) (0.0581) (0.0868) (0.0737) (0.0878) 2000 0.1010* 0.1100* 0.1149* 0.0892 0.1557* -0.0250 0.2043** (0.0573) (0.0575) (0.0639) (0.0587) (0.0901) (0.0801) (0.0882)


2001 0.1519*** 0.1595*** 0.1559** 0.1365** 0.1515 0.1034 0.2219** (0.0571) (0.0571) (0.0653) (0.0592) (0.0932) (0.0742) (0.0884) 2002 0.2592*** 0.2715*** 0.2695*** 0.2388*** 0.3219*** 0.1485** 0.3562*** (0.0549) (0.0552) (0.0668) (0.0583) (0.0884) (0.0680) (0.0863) 2003 0.3150*** 0.3261*** 0.3241*** 0.2863*** 0.3555*** 0.1555** 0.4221*** (0.0551) (0.0407) (0.0777) (0.0607) (0.0909) (0.0706) (0.0866) 2004 0.4532*** 0.4573*** 0.3907*** 0.4190*** 0.5323*** 0.3844*** 0.5416*** (0.0545) (0.0544) (0.1027) (0.0625) (0.0955) (0.0740) (0.0869) theta 3.51e-18 (2.07e-14) Wald chi2 1170.48 1288.92 49.11 1353.92 202.02 86.88 913.71 Observation 7287 7287 7287 7287 3605 2125 5162 Firms 2072 2072 2072 2072 958 521 1551 Log likelihood -34169.979 -34168.423 -34183.107 -34167.766 -14881.442 -8290.744 -22829.628

Note:

a. The key results of Table E are reported in Table 4. b. Variable BG* is predicted probability of adopting the Born Global internationalization process that is estimated from Table 3. c. Robust standard errors (corrected for clustering at the firm level) are in parentheses. d. ***, ** and * indicate statistical significant at the 1%, 5% and 10% levels, respectively.


born global, gradual global, and their determinants of

Documents